by Dave Goulson
In the mid-1990s a new class of insecticide was introduced. Known as neonicotinoids, they are synthetic variants of nicotine. They block open insect nerve-receptors, thus attacking the insect nervous system and brain, and are phenomenally toxic in tiny amounts. Of course insecticides wouldn’t be much use if they weren’t toxic to insects, so this might be regarded as a good thing. Neonics (they sound a little more friendly when abbreviated) have a major advantage over most of the insecticides that went before, in that they are systemic. They can be applied as a seed dressing before the crop is sown, and the germinating seedling absorbs the chemical, which spreads throughout the plant. Any herbivorous insect that eats any part of the crop dies. This is a wonderfully neat system. Previously insecticides had to be sprayed on to the crop from a tractor-mounted boom. Much of them landed on the soil, and in even a slight breeze they would blow in to the hedgerows. Only the parts of the crop that were directly coated with the spray were protected, so the lower leaves and roots would be vulnerable to herbivores. As the crop grew, further applications were needed on the new leaves. Overall, more chemical was needed to protect the crop, and the farmer had to spend both time and diesel applying it. Many of the chemicals used were pretty nasty – for example, the organophosphate insecticides were derivatives of nerve agents developed during the war to kill people – and so spraying them around posed a direct threat to the farm worker. All in all, it is easy to see why the neonics proved to be hugely popular, and they quickly became one of the most widely used classes of insecticide in the world. They now comprise about one-quarter of all insecticides used globally, and one type of neonic, known as imidacloprid, is the second most widely used agrochemical (after the herbicide glyphosate). In the UK, agricultural use of it has risen steadily, reaching about 80 tonnes per year at the last count.
The systemic nature of neonics is both their great strength and, perhaps, their Achilles heel. They spread to all parts of the plant, and that inevitably includes the nectar and pollen. If the crop is visited by pollinating insects, then they consume small amounts of these chemicals. Not long after the introduction of neonics in the 1990s, French beekeepers started to claim that they were causing their honeybee colonies to die. Their campaigns led to partial bans on some types of neonics on some flowering crops, but to the concern of French beekeepers, this largely meant that farmers simply used different types of neonics. This controversy rumbled on for some years until an incident in Germany in 2008. A batch of maize seeds had been coated with an incorrect formulation of neonic. The chemical was not properly stuck to the seed, and when the seeds were drilled, much of the coating blew away as a fine powder. Hundreds of honeybee hives in the area were wiped out more or less instantly. There was an uproar, and neonics were banned in Germany pending an investigation, but the ban was subsequently rescinded when it became clear that the problem lay primarily with the incorrect formulation. Nonetheless, this incident raised awareness among beekeepers and environmentalists as to the acute toxicity of these compounds to insects such as bees, and prompted further investigations.
Of course the chemical manufacturers were aware that neonics would get into the pollen and nectar of crops from the start. Agrochemicals go through various safety tests before they are licensed for use, including an evaluation of their toxicity to bees and other beneficial insects. Typically, groups of lab animals are fed on varying doses of a chemical and then monitored to see if they expire. This enables calculation of the ‘Lethal Dose 50%’ or LD50 – the dose that causes half of the test animals to die. The LD50s for the various different neonics in honeybees are all very low, just four- or five-billionths of a gram for the common types such as imidacloprid and clothianidin. To put that in context, one gram (not much more than the contents of a sachet of salt) is enough to give an LD50 to 250 million honeybees, or about twenty-five tonnes of bees.
How does the LD50 compare to the amounts found in nectar and pollen? Typically, the pollen of treated crops such as oilseed rape contains concentrations of neonics in a range from one to ten parts per billion – not much, but then these are very toxic chemicals. The amounts in nectar are usually even less, commonly below one part per billion. The big question then is: are these concentrations sufficient to harm bees? At ten parts per billion, a honeybee would need to consume about half a gram of pollen – about five times its own body weight – to receive an LD50. A bee would certainly not consume this much in a short period, although it could easily do so during its life. However, typical lab toxicity tests don’t look at long-term effects; most last just a couple of days, and over this period these sorts of concentrations do not kill bees. As a result, the compounds were deemed to be safe for pollinators, and licences were granted for their use all over the world. Whenever beekeepers claimed that these compounds posed a threat to their bees, the agrochemical companies pointed to the data and argued that the amounts bees consume are not enough to kill them.
The arrival of CCD provided new impetus to investigations into bee health, and a re-examination of the hazards that bees face in the modern agricultural landscape. It led to prominent campaigns by beekeepers and environmentalists around the world, many of them targeted at getting neonics banned. In the UK the invertebrate conservation charity Buglife produced a report on neonics that argued for a ban, supported by their counterpart in the USA, the Xerces Society. As someone involved in bumblebee conservation and research, I was regularly asked to support these campaigns, but I was reluctant to do so. I wasn’t aware of any compelling evidence that pinned either CCD or bumblebee decline to neonic use. It is the job of scientists, so far as is humanly possible, to be impartial and to provide the evidence that informs the decisions of others, not to become environmental lobbyists, although sometimes the distinction becomes blurred (such as when writing this book). Nonetheless this issue seemed to be one that wasn’t going to go away, so in 2011 I decided to do some research of my own.
As a first step, I tried to read everything that had been published already. This was hard to do, because the work done to examine the safety of these chemicals when they were first developed is not available for scrutiny by scientists or anyone else. Various summary reports could be downloaded from European and US regulatory agencies, but these rarely gave sufficient detail to understand fully what had been done. However, there were quite a few academic studies published in the mainstream scientific literature after the chemicals became widely used, mostly carried out in the lab or with bees flying in cages. Almost all agreed that, when exposed to realistic doses of neonics such as bees might encounter on a seed-treated crop, there was little or no mortality, at least in the short term. However, some found interesting effects on bee behaviour. Neonics are, after all, neurotoxins, so it seemed plausible that sub-lethal doses might adversely affect the behaviour of bees. Studies on both honeybees and bumblebees seemed to suggest that their ability to learn and to bring food back to the nest might be impaired when fed even minute amounts of neonics. However, the effects were generally small and nothing was seen that could explain the complete collapse of honeybee colonies.
I chatted over these studies with Penelope Whitehorn and Steph O’Connor, two members of my bumblebee research group. In nature, bees travel kilometres from their nest in search of patches of flowers. They have to learn how to get pollen and nectar out of the flowers (each flower being of a different design) and then find their way home. For their nest to thrive, each worker has to do this over and over again, all day long, for days on end. Their navigational abilities are amazing. They can use the sun as a compass; they seem also able to detect the Earth’s magnetic field; and they can memorise the position of various prominent landmarks such as trees and buildings. It seemed to us that what was really needed was a study of what happened to bees when exposed to neonics in a natural setting. If the exposure was impairing their mental faculties in some way, then the effects might not be at all obvious when the bees had to fly all of two metres from their nest to a dish of honey placed there fo
r them by the experimenter. Even a very poorly, intoxicated bee could probably manage that. On the other hand, anything that interfered with their navigation or learning would be much more likely to become a problem when faced with the challenges of the real world. If no effects were found even under such natural conditions, then I felt we might finally be able to stop worrying about neonics and look elsewhere for the cause of our bees’ problems.
We sat down and devised the best experiment that we could come up with. We had no funding for this, so it had to be simple and cheap. I persuaded a colleague named Felix Wackers, of Lancaster University, to provide us with native buff-tailed bumblebee nests for free, to which he had access through his association with one of the companies that rear bumblebees for commercial use. We wanted to simulate the situation in which a wild bumblebee nest finds itself near a field of oilseed rape that has been treated as a seed with imidacloprid. Oilseed rape flowers for about one month in spring, and at this time it is a magnet for bumblebees, so during flowering one might expect a lot of the nectar and pollen coming into any nest near a rape field to be from the rape itself. Ideally we would have placed our nests next to a treated and an untreated rape field and then compared the difference, but we could not find untreated fields and we had no funds to pay for them to be planted. In any case we would have had to be certain that there were no other treated crops within flight range of a worker bee, and given that they can easily fly a kilometre or two, this was never going to be a practical option. Instead we opted to expose the bees in the lab, and then put them out in the field. That way we could control exactly what they ate.
We fed one batch of nests on clean nectar and pollen, and another on nectar and pollen carefully mixed with imidacloprid, to re-create exactly the very low concentrations found in rape. After two weeks we took the nests out on to the campus at Stirling University and opened the doors. From that point on the bees were left to look after themselves; to gather food they would have to fly off and find flowers, just as they naturally would. We couldn’t be certain that they wouldn’t be exposed to more neonics in the gardens on the edge of Stirling, but no arable crops are grown nearby, so they weren’t likely to be exposed to much; and at least any differences between our treatment groups would have to be due to what they fed on in the lab before they went out.
Every two weeks Penelope and Steph went out in the middle of the night (when all the bees should be at home) to weigh the nests – one dark night scaring themselves half to death by imagining that a coil of hosepipe dangling from a wooden post was in fact a very tall and sinister man in a large floppy hat.
We analysed the data as they came in, eager to see if any differences emerged between the treated and untreated nests. Slowly the average weights of nests in the two groups diverged, with the untreated nests growing at a slightly faster rate. By six weeks the differences were quite marked. After eight weeks the nests were starting to senesce, losing weight and producing males and new queens as they naturally should in summer. We collected them in and dissected them, so that we could count exactly how many eggs, larvae, pupae and adult bees there were. The results were striking. In most respects the treated nests were just a little smaller, with fewer pupae and adult bees, but in the most important respect they were dramatically different. The control nests produced, on average, about thirteen new queens. The treated nests produced an average of just two – an 85 per cent reduction. These new queens are the only stage to survive the winter, and it is the queens that found new nests the following spring. All else being equal, an 85 per cent drop in queen production means 85 per cent fewer nests being founded the following spring.
Oilseed rape is a very common crop in lowland England, so few bee nests are far from a field of it. In a recent study of the Hertfordshire landscape we found that there was almost nowhere more than one kilometre from the nearest oilseed rape, which is easily within the range of foraging bumblebees. Oilseed rape attracts honeybees and a whole range of bumblebee species, particularly buff-tails, white-tails and red-tails, plus numerous hoverflies. Almost all of it is treated with neonics – I am told that it is near-impossible to get hold of untreated seed, even if a farmer wished to. These compounds are also routinely used as seed dressings on many other crops: sunflowers, sugar beet, potatoes, wheat and maize. Well over one million hectares of UK farmland are treated with them every year. Raspberries and strawberries are sprayed with neonics during the spring and summer, using much larger amounts that are used for seed dressings, and these crops are primarily pollinated by bumblebees. Garden insecticides are mostly based on neonics. For less than ten pounds your local garden centre will sell you a bottle of neonic containing sufficient active ingredient to kill instantly several million honeybees.3 These are advertised for use on flowers, and on flowering vegetables such as beans and peas. Unlike farmers, gardeners are entirely untrained in the use of pesticides, and most probably bung on a bit extra for luck. Bees in suburban areas are playing Russian roulette every time they feed on a new patch of flowers.
Neonics are also sold as soil drenches to kill subterranean grubs that eat grass roots in lawns, golf fairways and pastures; heaven forbid that a suburban lawn should have a few brown patches where the roots of the grass have been nibbled – far better that the whole lawn (and any clover or dandelion flowers it might produce) be impregnated with nerve toxins.
In urban areas trees are sometimes injected with neonics to protect them against pests; for example, avenues of trees in suburban streets may be treated to prevent outbreaks of aphids, which could result in unsightly and sticky honeydew on the cars below. An entire tree can be made toxic to all insect herbivores for several years to come by a single injection. If these trees happen to be lime trees, then bees will feed on them, adding yet more to the dose they receive.
Extrapolate our results across the country – indeed, across the world – and the likely scale of the impact on bumblebees is breathtaking and terrifying. One way or another, almost all bumblebee nests are likely to be exposed to these compounds. Perhaps we had discovered the ‘smoking gun’ at the root of bee declines. From being rather sceptical about the claims that neonics were wiping out bees, I found myself coming round to the view that this might well be true. We were, as you might imagine, very excited by this and keen to publish our work quickly and in a high-profile journal, where it would be noticed and acted upon. We submitted it to the journal Science, and waited with bated breath.
The peer-review process for scientific publications can be frustratingly slow, and it was many weeks before we got a reply. Even then it was not clear whether Science would eventually publish our work, for the anonymous referees to whom they had sent our paper had recommended various changes, and one of them didn’t seem to think the work was particularly interesting. We did our best to comply, returned the manuscript and waited once more. Eventually, to our huge relief, Science declared that it would publish the work. It also revealed that it intended to publish a second, related paper alongside our own. I begged a copy and was fascinated.
It seemed that a French team, based in a government lab in Avignon, had also decided to conduct more realistic experiments on the impacts of neonics on free-flying bees. They had studied honeybees, exposing foraging workers to tiny doses of a neonic called thiamethoxam, mimicking them discovering and feeding upon a treated crop. They had glued miniature radio tags to their bees, so that the return to their hive was automatically detected and recorded by sensors mounted on the hive. Their results provided a beautifully simple explanation for the slow growth and poor performance of our treated bumblebee nests. They found that worker honeybees were much more likely to get lost on the way home if they were fed a neonic. The effects were more pronounced the further the bee was from home, and if the bee was in an unfamiliar location, from which finding its way back to the hive would require its navigation skills to be in tip-top condition. In my mind there is a simple human parallel: it is easy for a drunk to find his way home from his favourite pub
, particularly if it is close to home, but put him in an unfamiliar pub and he is quite likely to get lost. The French team’s findings provided the first clear indication of a mechanism that could explain the symptoms of CCD. CCD is not about bees dying; it is about them disappearing. If bees cannot find their way home, then a hive will quickly empty of bees, leaving no corpses behind. Lost bees are as good as dead. Without their hive they have no purpose in life and will quickly expire. So it seems that sub-lethal doses of neonics can indirectly kill bees in the real world, while having no measurable effect in the lab.